The pollution sources identification model is presented by the coupling of the river water quality model and particle swarm optimization (PSO) algorithm to estimate pollution sources from the measured/simulated contaminant concentration in the river. The “twin experiment” is adopted to verify the feasibility of the identification model, and three cases are constructed to explore the results of the identification model in different situations. The experiment test demonstrated that the identification model is effective and efficient, while the model can accurately figure out the quantities of the pollutants and position of a single pollution source or multiple sources, with the relative error of the mean is less than 3%. Many factors are explored, including the level of random disturbance and the impact of particle population size. The outcome showed that the disturbance level is less than 5%, thus the precision is preferable, and when the number of particles is three, the identification is the best. When performing multiple sources, identification of multiple sets of monitoring sections respectively can obtain more accurate results with less error. In this paper, the optimization method of the inverse problem is applied to the identification of river pollution sources, which can help us to identify pollution sources and provide us a scientific basis for subsequent water pollution control and prevention.

污染源识别模型是通过河流水质模型和粒子群优化（PSO）算法的耦合提出的，从河流中测量/模拟的污染物浓度估计污染源。采用“孪生实验”来验证辨识模型的可行性，并构建三个案例来探讨辨识模型在不同情况下的结果。实验测试表明，该识别模型是有效且高效的，该模型能够准确地计算出单个或多个污染源的污染物数量和位置，均值相对误差小于3%。探索了许多因素，包括随机扰动的水平和粒子群大小的影响。结果表明，扰动水平小于5%，精度较好，当粒子数为3时，识别效果最好。在执行多源时，分别识别多组监测断面可以获得更准确、误差更小的结果。本文将逆问题的优化方法应用于河流污染源的识别,可以帮助我们识别污染源,为后续的水污染防治提供科学依据。分别识别多组监测断面，可以得到更准确的结果，误差更小。本文将逆问题的优化方法应用于河流污染源的识别,可以帮助我们识别污染源,为后续的水污染防治提供科学依据。分别识别多组监测断面，可以得到更准确的结果，误差更小。本文将逆问题的优化方法应用于河流污染源的识别,可以帮助我们识别污染源,为后续的水污染防治提供科学依据。